Handle Azure NIC reservation during VM deletion

[apoorvajagtap]

Issue: rancher/rancher#47903

Azure may temporarily retain NIC reservations after VM deletion or failed VM provisioning, causing cleanup to fail with the NicReservedForAnotherVm error.

This commit:

• Implements exponential backoff retry for reserved NIC deletion (2s initial, capped at 180s total).
• Handles context cancellation during wait periods

This improves cleanup reliability by preventing orphaned NICs and, consequently, orphaned Public IPs.

Test Validation:

• Please refer To Reproduce section on the issue.

[apoorvajagtap]

I think proactively detaching NICs from VMs during cleanup could be a more robust approach, as it would help avoid orphaned resources and also allow for better parallelization. If this sounds like a preferable alternative, I’d be happy to explore its feasibility. (ref)

[jakefhyde]

1 nit

[Copilot]

Pull request overview

This PR improves Azure resource cleanup reliability by handling the transient NicReservedForAnotherVm condition that can occur after VM deletion or failed provisioning, which otherwise can leave orphaned NICs (and consequently Public IPs).

Changes:

• Adds a maxRetryDuration constant (180s) used to wait out Azure’s NIC reservation window.
• Updates DeleteNetworkInterfaceIfExists to detect the NIC-reserved error, wait, and retry NIC deletion once, while respecting context cancellation.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[jiaqiluo]

Left one comment. Also, please review the Copilot comments.

FYI, this repo is not branched for Rancher release lines, so you might need to backport the changes to other lines.